B2B Customer Segmentation: Monthly Recurring Revenue and Cost Analyses

Last week, I explained how a census composition comparison analysis can be used to quickly identify potential B2B customer segmentation characteristics and/or inform segmentation decisions. In that post, I also shared a step-by-step guide on how to set-up and execute these types of analysis.
This week, I will explain how a combination of monthly recurring revenue (MRR) and monthly recurring cost (MRC) analyses of your customers can be used to identify a target segment.
The purpose of Monthly Recurring Revenue and Monthly Recurring Cost customer analyses are to identify customer characteristics that are correlated with strong revenue and lower cost of support and service, respectively.
These analyses can be done rather quickly if you focus on the characteristics that you already track about your current customers and limit the additional secondary data collection to no more than a couple easy to collect data points. The characteristics you will commonly want to be able to look at are company size, sector, customer type (B2B, B2C, B2B2C), use case, and geography.
The most time-intensive aspects of this analysis are mapping costs of service back to your customers (if this hasn’t been done already) and collecting additional data that is not incorporated in your current customer data. I recommend running the analysis across the key factors you do have data on first to see if any of them highlight potential segmentation lines.
There are many different ways that you can conduct Monthly Recurring Revenue and Monthly Recurring Cost customer analyses. Below are step-by-step guides on how to conduct these types of analyses using quintiles and normalized averages per a deal. These approaches can be used in tandem or as separate analyses. Using them in tandem is more time consuming, but it gives you a more complete view of potential customer segmentation schemes.

Quintile Approach

1. Identify your best customer characteristic data source(s) and combine them together if necessary. This will generally be your CRM.
2. Calculate the monthly recurring revenue for a customer. To get this number just subtract one-time customer set-up costs from the overall revenue and divide it by the total months under contract. The best source for this data point is generally the transactional database.
3. Calculate the monthly recurring cost of service and support for each customer. This may require going into the companies support records and allocating costs based on time spent working with each customer. This figure should be calculated on a per-month basis. The customer tenure used in the monthly recurring revenue calculation can also be used to get a per-month statistic. Most companies regardless of size will track this to some degree. Long-term, a company will want to be able to allocate these costs for specific contact reasons. For more on this, see my blog post on how to use contact codes.
4. Append Monthly Recurring Revenue and Monthly Recurring Cost data to the relevant customer characteristics data.
5. Now you will want to determine which quintile each customer belongs to for the monthly recurring revenue and monthly recurring cost variables. This is a way of standardizing the data so that you can try to identify which customer characteristics are most likely to lead to a customer with strong revenue and manageable costs.
6. To do so, you will want to run counts across each of the segmentation characteristics to see which characteristics enable you to identify the best set of customers in terms of monthly recurring revenue and monthly recurring cost. 
7. Now you will want to chart the quintile counts across each of the segmentation characteristics you are testing to see if you can identify any patterns. Often, the best segmentation criteria are actually developed by looking at trends across one or multiple segmentation criteria. For example, you may find that a series of related sectors all look good and this may suggest to you that a certain type of business model or business is a good target.
8. After analyzing the customers across each of these cuts, you should have a pretty good idea which of these variables has the most explanatory power when it comes to cost or revenue. You will want to look for positive and negative relationships. Knowing what is bad is almost as useful as knowing what is good — it can help you prioritize targets.

Normalized Averages Per a Deal Approach

1. Repeat steps 1 – 4 above if you are using the normalized average approach on its own. If not, you will want to start with the same core data set as you used with the previous analysis exercises.
2. Look at the distribution of the monthly recurring revenue and monthly recurring cost variables to see if there are any major outliers. To do so you will want to calculate some summary statistics about the distribution, including standard deviation, min, max, mean, and median, as well as how many standard deviations the min and max are away from the mean. This will also help you identify the major outliers in the data and help you get a sense as to what percentage of the data should be removed from the top and bottom of the data to normalize it. You will want to do this for each variable irrespective of the other.
3. Calculate the normalized average monthly recurring revenue and average monthly recurring cost per a customer across each segmentation characteristic of interest (i.e. sector, company size range, business model, etc.).
4. Next you will want to chart the normalized averages per-deal for each of these statistics across each segmentation criteria, and analyze the results for patterns just as you did with the quintile approach. You will want to look for segments that have strong revenue values per a deal and low costs.  You will want to look for positive and negative relationships with revenue and costs.

You can also do this using regression, but that is a subject for another blog post. I recommend reading the OpenView eBook, Finding Your Best Customer: A Guide to Best Current B2B Customer Segmentation, if you are looking for more information on that topic.
Next week, I will explain how to set-up and execute a customer tenure analysis for the purpose of segmentation.